The analysis of a parameter identification algorithm which was derived from the continuous time Kalman filter

The matrix Riccati equation associated with the identification algorithm may be regarded as two sets of scalar equations representing the squares and products of pairs of errors in the estimates of the parameters. Although the initial condition for the Riccati equation is arbitrary, it is shown that its elements tend to the ratio of the product of the actual estimation errors. This method of identification is shown to be stable but rather sensitive to numerical errors in its computation. However, an outcome of the analysis is that an alternative algorithm, which eliminates the Riccati equation, is suggested. Related convergence and stability properties of state observation by the deterministic filter are also discussed.